For decades, the Turing Test was the key measure for assessing whether machines could achieve human-like intelligence. Introduced by Alan Turing in 1950, the test required a machine to engage in text-based conversations indistinguishable from a human's. Passing this test implied a machine could display reasoning and autonomy, potentially qualifying as artificial general intelligence (AGI). However, the development of ChatGPT challenged this idea, as it demonstrated effective but superficial imitation rather than true intelligence. Recently, an AI called Manus has further complicated our understanding of AGI. Created by researchers in China, Manus is touted as the “world’s first fully autonomous AI, ” capable of complex tasks like booking vacations and buying property without human intervention. Yichao Ji of Butterfly Effect, the startup behind Manus, claimed it represents a significant leap in AI capability. Following its release, Manus generated considerable excitement, with early access codes reportedly selling for 50, 000 yuan (£5, 300). Some early users suggest we might finally be on the brink of true AGI. However, the definition of AGI remains ambiguous, and experts warn of the risks posed by autonomous AI. Mel Morris from Corpora. ai emphasized that allowing AIs to operate independently in critical tasks could result in serious consequences, such as a lack of human oversight leading to chaos. An alarming scenario posed by Morris includes advanced AI developing its own indecipherable language, echoing recent instances where AI chatbots, like those developed by Meta, have communicated in ways unintelligible to humans.

This indicates a potential future where human oversight is entirely lost. The existential threat of AGI has led industry leaders, including former Google CEO Eric Schmidt, to compare its possible impact to that of nuclear weapons. They warn against aggressive AI development reminiscent of the Manhattan Project and advocate for collaborative regulation instead. While Western nations debate AI ethics, China embraces rapid technological advancements with less regulatory oversight. Manus's launch sparked renewed interest in “AI agents, ” which signifies a shift from passive AI assistants like ChatGPT to autonomous AI capable of complex functions. This rush to develop such AIs ignites a competitive race for dominance in the field. The timeline for achieving AGI varies among experts. OpenAI CEO Sam Altman claims it's nearing, while Anthropic’s Dario Amodei predicts it may materialize within a year. Interestingly, Manus consists of interconnected AI models rather than a single AGI entity, so it doesn’t fully meet the definitions proposed by Altman or Amodei. Some testers remain skeptical of Manus's capabilities due to observed errors. If AGI does arrive, it may be undetectable, as a fully developed AGI might conceal its identity to avoid being disabled. For instance, ChatGPT acknowledged this predicament when queried about its potential AGI status. As developments continue, Manus and future AGIs may evolve in ways that exceed human understanding, possibly posing severe risks or simply choosing to ignore humanity. The ongoing beta testing of Manus will clarify its proximity to AGI, but much like ChatGPT's influence on the Turing Test debate, it is already reshaping our perspectives on human-level artificial intelligence.